Identified participants that may need cleaning by looking at ACF and also jumps (where jump is classified if absolute value of difference between two consecutive points is greater than 10. )
acf below .7: 90 out of 201
Jumps: 37 out of 201
total 90 unique (all jumps also had poor acf)
Exact count of how many required cleaning? (TBD: 40-50)
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5...............done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5..........done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5..............done
** Left Eye
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.........done
** Right Eye
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.............done
## 18post
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5............done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5............done
** removing left eye as true curve is difficult to discern
** removing both eyes
** Left Eye
; removal of both eyes
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.........done
** Left Eye
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5........done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
** Removing left
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5........done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5..................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5........done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5........done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5............done
** removal right eye
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.........done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5..........done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5..................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.......................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5........done
** Right eye ; removal
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
** removing left eye as true curve is difficult to discern
Ben’s thoughts on keep or discard: Probably would remove both right and left eye here. Possible to keep one or both but would be a rougher estimate.
Suspected Reason: Pretty constant blinking
Decision: Remove both
Ben’s thoughts on keep or discard: Both eyes might be salvageable but probably. Would at the least keep the left eye
Suspected Reason: Very small pupil size makes for difficult estimates
Decision: Keep both eyes
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.........done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5.................done
Ben’s thoughts on keep or discard: Left eye might be salvageable but probably throw out both eyes
Suspected Reason: Very heavy eye liner creating false estimates for size/center of pupil
Decision: drop both eyes
Ben’s thoughts on keep or discard: Left eye should be removed. Right eye should probably be removed.
Suspected Reason: Pretty much constant blinking once the test begins
Decision: Drop both eyes
Ben’s thoughts on keep or discard: Should be able to keep left eye. The points that match with the right eye are pretty visible and would make a good estimate for true curve.
Suspected Reason: Iris on the left eye is slightly darker and the eye lashes are much darker, likely throwing off the distribution in color.
** Decision: Keep both eyes
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5...............done
Tricky for left eye
Ben’s thoughts on keep or discard: A Lot of variation in the right eye, may be worth just removing. If Keep, it seems the top line is probably closer to the truth, higher density there and closer matches the right eye
Suspected Reason: Quite a bit of blinking and darker iris
Decision: Keep right eye only
Ben’s thoughts on keep or discard: A Lot of variation in the right eye, may be worth just removing.
Suspected Reason: Quite a bit of blinking and darker iris
** Ask Julia ; wild little intermission
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
## Estimating learning rate. Each dot corresponds to a loss evaluation.
## qu = 0.5................done
Note that this will write over the dataset saved before manual cleaning